AI: The Shiny New Object in Search
It’s apparently a new day in Search. Why? AI (Artificial Intelligence) is officially integrated into two notable search engines, Google and (the new) Bing.
I finally had a chance to digest some of the headlines and read a few industry articles on the latest shiny new object in SEO: AI in Search. Here’s an overview of what I think is interesting and will be the trends to watch.
Interestingly, in Google’s attempt to beat Bing to the punch announcing their AI product it came off as the product being rushed to market and the presentation itself wasn’t as well coordinated as it should have been resulting in the company stock tanking.
Alphabet lost about $100 billion in value after a demo meant to show off the AI-powered chatbot bungled its response on the NASA telescope.https://www.cnet.com
That is not a headline any investor wants to wake up to.
For context, Google announced on February 6th the introduction of Bard to Search. At first, based on the location of AI in their SERP, it appeared to be a revamped version of a Featured Snippet, an organic feature that sits at the top of the results page.
What is the purpose of artificial intelligence?
Google’s product announcement made me wonder what queries they were trying to target. What problem were they trying to solve? Reading through their announcement, though, I think this key concept completely flew under the radar:
“AI can be helpful in these moments, synthesizing insights for questions where there’s no one right answer. Soon, you’ll see AI-powered features in Search that distill complex information and multiple perspectives into easy-to-digest formats, so you can quickly understand the big picture and learn more from the web”Google AI Search announcement
Still, it wasn’t until I read Brody Clarke’s article where he talked about “NORA (No One Right Answer)” that I realized incorporating AI in search is trying to solve for informational searches with no one right answer.
What is artificial intelligence in simple words? It’s an Ask-Me-Anything machine.
A tale of two products – Google Bard & Bing AI Chat
Google Bard: Citations – Out, damned spot; out, I say! —One, two: why then, ’tis time to do’t.
Bard…Shakespeare. Get it?
The key drawback of how AI is being presented, at least initially in Google’s interface, is that the AI responses do not provide clear links or citations to web publishers. Glenn Gable’s article points this out as “Google’s war against publishers” and he’s not wrong. But I think the behavior we’ve seen with Featured Snippets tells a similar story because there was a similar argument with the emergence of zero click search results: “if I can see the answer in the SERP, why do I need to click on the result?” Everyone was worried about their organic traffic and CTR when zero click results emerged.
Over time, earning the Featured Snippet spot did result in some organic traffic as those consumers who desired to learn more clicked through. Gabe added, “It’s also worth noting that Google has not answered any questions about
citing sources. And I mean literally nothing has come from anyone at Google about linking to publishers (which makes me think they were unprepared for the question). That’s also scary…so we’ll see where this ends up.”
That should have been a core tenant of the product. It’s very strange that no one commented on citations.
On a separate but related note, what’s come along with the announcement of AI driven content is speculation that people think their jobs will be taken away because AI will create all of this incredible content at scale. But that’s a fallacy.
AI responds to prompts from humans and can only learn from what humans teach it or feed it. It can’t think for itself. And it’s prone to inaccuracies and biases. AI generated content needs and will continue to need human oversight.
Lily Ray makes a solid point about the delay with information used to train AI content generation tools in her article:
“For example, ChatGPT was trained on data ending in late 2021, although the tool does appear to be improving to reflect more recent information. Given the importance of fresh information in so many areas of SEO, this is a significant limitation for the tools to be able to produce entirely helpful content.”
In relation to the SEO community, the guidance Google emphasized about content creation was that it comes down to the quality of the product. AI can be involved in generating content (there’s no penalty for this), but if it isn’t helpful to users, it will get weeded out by the systems & algorithms in place (i.e. Helpful Content Update).
Marie Hayne’s newsletter highlights this as well: “It is perfectly fine to use AI in your content creation efforts as long as your end product demonstrates E-E-A-T and is helpful to people. However, Google does not recommend you list AI as the author of your content.”
The extra “E” is for Experience.
Bing AI Chat contains citations (peasants rejoice!)
The next day, Microsoft had a huge event in Seattle where they showcased their AI Search product. Their implementation was well thought out and wow’ed the media. Bing’s AI chat attributed the information to sites where it “learned” about the answers it was providing.
This is a step in the right direction in terms of product attributes but, without the volume of users Google can drive, it’s small potatoes to publishers.
What’s interesting about the approach Bing is taking, in their POV, is that it’s time for a new approach. Barry Schwartz’s attended the announcement and posted about it on Twitter noting:
- Bing will launch an all new search engine with AI powered. It’s better, will answer questions, you can chat with it and it can create content for you.
- 40% of all queries result in someone clicking back and most searches are three keywords or less…
- he (Yusuf Mehdi) explains navigational, informational and other searches people do today. 50,000 people’s searches go unanswered, which is why it is time for a new approach with search.
Product War: To cite or not to cite
Whether or not annotations are visible in the SERP, IMHO, is irrelevant. The important piece for publishers is being rewarded with organic traffic when the AI result actually clicks through to the publisher’s site. Regardless of whether it’s a Google or Bing search, if the No One Right Answer section from AI doesn’t link directly to the original source(s), that’s when you’ll really have an uprising on your hands.
At that point there’s zero incentive for publishers to invest time and resources creating content if the level of organic traffic as a referral source in your analytics doesn’t deliver.
Speaking of citing sources, what if there’s a scenario where you don’t want AI chat bots learning on your content i.e. subsequently misquoting or using it out of context? I thought that was an interesting consideration seeing this article in the SEO FOMO newsletter by Aleyda Solis: How to Block ChatGPT from learning on your website, by Roger Monti
The technique involves utilizing the Robots.txt but it’s not a guarantee. At the very least, it’s nice to have an idea of how to attempt to do this if such a scenario presents itself.
Aleyda always does a great job curating her newsletter content. I highly recommend signing up.
AI trends to watch
- Product dominance will win.
Will Bing’s AI product tip the scales in their favor where consumer behavior will change and more users will start with Bing vs. Google? IMO, I don’t yet
think this is likely for a 10-20 years because, as Gabe pointed out, “Google typically drives exponentially more traffic than Bing” (his post offered a handful of comparisons showing the difference between GSC and BWT Clicks & Impressions data). The adoption curve is still too great.
- Being second best in the market means more room to test & iterate.
Microsoft has multiple sources of revenue and Search is a relatively small one at this point. They have the advantage at this point because they can afford to “test and learn” their way from the product they launched this month to a more refined, mature product. Whereas Google has to worry about its impact to Search; bigger changes are riskier because they have the lion’s share of the market.
- There will be short term & long term effects of AI – and they’ll be very different.
Gabe’s idea that AI is headed towards manifesting “Jarvis for Search” (as in the AI assistant created by Tony Stark) got me thinking. I think it would be incredible because Jarvis is Stark’s right hand and intuitively knows what,
when and how he needs answers or action. But I’d only be on board if it operated in a closed network; can you imagine your Jarvis inundating you with advertising or spam?
IMHO, it’s one thing for Bing to say it wants to provide links to publishers in their AI chat.
But the bigger question is “will more people use Bing as opposed to Google?”
User behavior is a completely different animal; influencing it…changing it…creating new habits…that takes a long time. In 2023, Bing is not the current market leader when it comes to search engines. Google is. And by a very large margin.
Hypothetically, even if Bing has a superior search product with the addition of AI, it’s still all about adoption and preference and doing what’s easiest.
The consumer POV is: “What app is on my phone that I feel gives me the fastest, most reliable, accurate results?” That wins every time.
As far as No One Right Answer results go, take it with a grain of salt. These AI products were basically rushed to market and there’s a long way to go towards improving their outputs.
The startling news just keeps coming:
- How Googlers were told to fix responses from Bard, its AI chatbot
- Bing AI Can’t Be Trusted, blog article by Dmitri Brereton
For the record, none of this was written using AI. I used my brain, my experience as an SEO professional and a keyboard to write out my thoughts.
The information contained in this post does not reflect the views of my employer.